,
6 tweets,
2 min read
Read on Twitter
Re polling: Pollsters split on poll aggregators' use of data, but do caution 'buyer beware' for the public - The Hill Times hilltimes.com/2019/07/29/pol… #cdnpoli
Tips for reading polls:
(1) Polls are a reflection not of what's going to happen, but of what has already happened.
(2) The further away a poll is done from the event it's about (for ex, the Oct. 21 federal election) the less it should be taken as a predictor of that election
(1) Polls are a reflection not of what's going to happen, but of what has already happened.
(2) The further away a poll is done from the event it's about (for ex, the Oct. 21 federal election) the less it should be taken as a predictor of that election
(3) It's much harder for a pollster to determine seat results from national polling, particularly almost three months before the federal election.
(4) Pollsters findings can be dramatically different from one another because they use different methodologies.
(4) Pollsters findings can be dramatically different from one another because they use different methodologies.
(5) While there are always a few 'rogue' polls leading up to an election, by the time of the actual election, most pollsters are in the ballpark of what's going to happen on election day, particularly those who poll right up to before the election.
(6) Finally re Trump, most pollsters in the US correctly predicted that Clinton would win more of the popular vote than Trump. What was astounding is that very few of them factored in that the president isn't elected by the popular vote but by the electoral college.
That means that an American president can win the election with fewer votes than his or her opponent, depending on his or her respective strength in each of the 50 US states.
